With security being a hot topic in society, a use case related to experimenting with 5G for a safe urban environment was a logical outcome. A team of scientists, Phds and 5G experts conducted research with Spot, the robotic dog. The aim: to investigate how Spot can be of added value when a bag is left abandoned. Meet the future neighbourhood watch…
Several municipalities in the region Rotterdam – The Hague have shown great interest in the possibility of additional services by a robot dog. The research consisted of several components. First of all, to get Spot functioning correctly with a 5G network. Then, taking the necessary steps to detect and pick up a bag. And finally, to investigate how the public reacts when confronted with Spot. Maneesh Kumar Verma, who is involved in all kinds of research with Spot, explains how the team overcame challenges and what they learned from this research.
City-like environment
“It’s really convenient that the 5G installation is located on TU Delft campus”, says Maneesh. “We can experiment in a city-like environment, without drawing a lot of attention from the public. The reason why we needed 5G was that we have LIDAR, but transmitting such a large amount of data with Wi-Fi 5 was not possible. You need a lot of computational power to operate the robot. You constantly ping data back and forth, as the robot interacts with the environment. Therefore, you need extremely low latency and extreme high reliability”.
Once having established the 5G connection, which was an important hurdle as very few people knew how this new technology should be installed, there was still a lot of manual programming to be done.
“As we use a variety of sensors from different companies, we had to configure everything again to work on the network. This was a huge learning experience for the students involved. They had to learn most things from scratch. The sensor box has to detect a bag, calculate the position, calculate how to navigate Spot towards the bag, how to find it’s way, pick up the bag, and so on and so forth. That took a lot of time and effort. And students don’t work full time of course,” laughs Maneesh.
Suspicious activity
Finally, when Spot was ready to go, the team could start to define when a bag is actually suspicious. “People leave their bags on the ground constantly. So when is the distance big enough between a bag and a person standing in the vicinity? And can we describe what suspicious activity is, say, in a train station? These were interesting discussions”, states Maneesh. “Cameras and AI can recognise a potential threat. This is where Spot takes action. We initially remotely control Spot to walk towards the bag, pick it up and bring it to a desired location. For instance, a separate room, away from a crowded area, where the bag can be investigated by a professional”.
This sounds pretty easy, but even with a remote control, Spot has to be configured in the right way, in order to function autonomously in the future. His sensors should detect a bag, even when there is hardly any contrast between the colour of the bag and the background. Furthermore, Spot needs to learn how to pick up the bag. With enough caution, to not harm the contents. “But then again, during tests with a bag that contained a laptop, the bag’s fabric was a bit smooth and the bag dropped on the ground. So the grip must be tight enough to prevent that”, describes Maneesh. Another challenge for Spot: lots of passers by. “When spots needs to move forward and there is an object or person in front of it, it stops its movement. From a safety point of view, we want to prevent a collision. But when people move around it constantly, it looks like Spot freaks out. Attempting to move, stop, move, stop, and so on. This makes people feel uncomfortable, because the movements look hesitant and raise doubt. We definitely need to work on that.”
Speaking of the public’s reaction to Spot, the research also took this aspect seriously. During multiple tests at Rotterdam’s main train station, they took surveys. In the future, it will be very important that Spot doesn’t raise alarm or create panic in the public space. Interviewing all kinds of travellers gave a few insights. Maneesh shares the first overall trends: “In Delft, people are used to high tech experiments and are not shocked to see a robot dog. That’s why we chose Rotterdam. Most people are surprisingly comfortable with it, as long as it’s monitored. In general, people over 40 are less comfortable with Spot. Kids on the other hand, adopt it very easily. The current young generation is growing up with a lot more automated devices than their parents. In time to come, audience will get used to these kinds of robots and technologies. You don’t get scared by a robot lawn mower or vacuum cleaner, right? These automated devices will be expanded. I envision that it will like an ambulance or police car passing by: we make way so it can go where it needs to go. We need to educate people that the robot dog needs to check something out.”
Characteristics
In order to get there, a lot of additional research is needed. Making the system robust was the first key goal. Next year, Maneesh and his team want to achieve the automation of many more processes through the 5G network. Furthermore, they will perform more human-machine interaction, to train and enhance Spot’s movements. There is also a bit of an issue with the design: “the paws can get slippery when they get wet, which is not ideal in The Netherlands. So we need to create more grip.” And last but not least, the team will research which characteristics could increase the public’s acceptance. Maneesh cites a few examples: “Think of the robot’s dimensions or the colour and print. We can advise suppliers about these appearance features, in order to have the robot dog perceived as a normal thing in fifteen years from now. The main obstacle will probably be the legal hurdles. But our partners at Erasmus university are already studying on that topic, so I am anxious to see what the outcome will be.”